生成對抗網路(Generative Adversarial Networks, GAN)其目的是透過生成網路(generative network, GN)與鑑別網路(discriminating network, DN)的對抗方式，產生出幾可亂真的虛擬圖片。為了改善GAN生成樣本的耗時問題，後續研究將Wasserstein距離導入GAN而提出了WGAN。然而無論是GAN或WGAN，皆從潛在空間(latent space)隨機取樣作為GN的輸入，因此使得GN需經過長時間的訓練方能產生可通過DN辨識的虛擬樣本。為進一步改善此耗時的問題，於數值分析領域，本研究提出一個模組化(Modularized)的WGAN，稱為MWGAN。MWGAN透過特徵選取法找出與類別標籤具有高度相關的關鍵屬性；基於關鍵屬性使用分群方法釐清其值域分布；非關鍵屬性，則基於此分群結果，以盒鬚圖推估其值域合理範圍，做為潛在空間隨機取樣範圍之依據。於MWGAN訓練過程中，GN的輸入項將分成常數以及變量兩類，其中常數為輸入樣本關鍵屬性之值，變量則為非關鍵屬性，其值將於設定的潛在空間隨機取得，因此透過MWGAN產生之虛擬樣本，關鍵屬性值與原始樣本相同，不同者僅非關鍵屬性值。本研究透過生成之虛擬樣本，重建非相關屬性與分類標籤之關聯，進而使建模工具從中找出更多非關鍵屬性之資訊，以做為決策者之參考。

Generative Adversarial Networks (GAN) was proposed in recent years to improve effectiveness of image generation. Through the generative network (GN) compete with discriminating network (DN), GAN could generate the picture which cannot be distinguish by people. GAN’s inputs is random select from latent space, thus it must spend a lot of time to generative image. To solve this problem, newer study imports Wasserstein distance to GAN and proposed WGAN. To speed up GAN’s consuming several times in numeric data aspect, this study proposed a modularized WGAN, named MWGAN. MWGAN use feature selection to find key attributes means high correlation between attributes with labels. Using clustering algorithm to classify the key attribute’s range, non-key attribute map to clustering result and using box plot finding its range. During the practicing period, GN’s input separate to constant and variable two parts. Key attributes are constant part input, non-key attributes are variable part input. The non-key attributes are random selecting from its range. Thus, MWGAN generates virtual sample which key attribute value are same as original sample, only non-key attributes are different from original sample. Through virtual samples generating by MWGAN, reconstruct relation between non-key attribute and label, helping modeling tools to discover the information hiding in the non-key attributes.

葉怡成，類神經網路模式應用與實作，儒林圖書公司，民國九十二年三月。
Arjovsky, M., Chintala, S., & Bottou, L. (2017). Wasserstein gan. arXiv preprint arXiv:1701.07875.
Ester, M., Kriegel, H.-P., Sander, J., & Xu, X. (1996). A density-based algorithm for discovering clusters in large spatial databases with noise. Kdd, 96(34) 226-231.
Faloutsos, C., & Kamel, I. (1994). Beyond uniformity and independence: Analysis of R-trees using the concept of fractal dimension. Paper presented at the Proceedings of the thirteenth ACM SIGACT-SIGMOD-SIGART symposium on Principles of database systems.
Fiore, U., De Santis, A., Perla, F., Zanetti, P., & Palmieri, F. (2019). Using generative adversarial networks for improving classification effectiveness in credit card fraud detection. Information Sciences, 479, 448-455.
Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., Bengio, Y. (2014). Generative adversarial nets. Advances in Neural Information Processing Systems, 27, 2672-2680.
Gulrajani, I., Ahmed, F., Arjovsky, M., Dumoulin, V., & Courville, A. (2017). Improved training of wasserstein GANs. Advances in Neural Information Processing Systems, 30, 5768-5778.
Huang, C. F. (1997). Principle of information diffusion. Fuzzy sets and Systems, 91(1), 69-90.
Huang, C., & Moraga, C. (2004). A diffusion-neural-network for learning from small samples. International Journal of Approximate Reasoning, 35(2), 137-161.
Jang, J.-S. (1993). ANFIS: adaptive-network-based fuzzy inference system. IEEE Transactions on Systems, Man, and Cybernetics, 23(3), 665-685.
John, G. H., Kohavi, R., & Pfleger, K. (1994). Irrelevant features and the subset selection problem. Machine Learning Proceedings 1994, 121-129.
Li, D.-C., Lin, W.-K., Chen, C.-C., Chen, H.-Y., & Lin, L.-S. (2018). Rebuilding sample distributions for small dataset learning. Decision Support Systems, 105, 66-76.
Li, D.-C., & Lin, Y.-S. (2006). Using virtual sample generation to build up management knowledge in the early manufacturing stages. European Journal of Operational Research, 175(1), 413-434.
Li, D.-C., & Liu, C.-W. (2010). Extending attribute information for small data set classification. IEEE Transactions on Knowledge and Data Engineering, 24(3), 452-464.
Li, D.-C., Wu, C., & Chang, F. M. (2005). Using data-fuzzification technology in small data set learning to improve FMS scheduling accuracy. The International Journal of Advanced Manufacturing Technology, 27(3-4), 321-328.
Li, D.-C., Wu, C.-S., Tsai, T.-I., & Lina, Y.-S. (2007). Using mega-trend-diffusion and artificial samples in small data set learning for early flexible manufacturing system scheduling knowledge. Computers & Operations Research, 34(4), 966-982.
Liu, Y., Zhou, Y., Liu, X., Dong, F., Wang, C., & Wang, Z. (2019). Wasserstein GAN-based small-sample augmentation for new-generation artificial intelligence: a case study of cancer-staging data in biology. Engineering, 5(1), 156-163.
MacQueen, J. (1967). Some methods for classification and analysis of multivariate observations. In Proceedings of the fifth Berkeley symposium on mathematical statistics and probability (Vol. 1, No. 14, pp. 281-297).
Mirza, M., & Osindero, S. (2014). Conditional generative adversarial nets. arXiv preprint arXiv:1411.1784.
Rousseeuw, P. J. (1987). Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. Journal of Computational and Applied mathematics, 20, 53-65.
Sannidhan, M. S., Ananth Prabhu, G., Robbins, D. E., & Shasky, C. (2019). Evaluating the performance of face sketch generation using generative adversarial networks. Pattern Recognition Letters, 128, 452-458.
Schubert, E., Sander, J., Ester, M., Kriegel, H. P., & Xu, X. (2017). DBSCAN revisited, revisited: why and how you should (still) use DBSCAN. ACM Transactions on Database Systems (TODS), 42(3), 1-21.
Spanhol, F. A., Oliveira, L. S., Petitjean, C., & Heutte, L. (2016, 24-29 July 2016). Breast cancer histopathological image classification using Convolutional Neural Networks. Paper presented at the 2016 International Joint Conference on Neural Networks (IJCNN).
Tang, D., Zhu, N., Yu, F., Chen, W., & Tang, T. (2014). A novel sparse representation method based on virtual samples for face recognition. Neural computing and applications, 24(3), 513-519.
Thaseen, I. S., & Kumar, C. A. (2016). Intrusion detection model using chi square feature selection and modified Naïve Bayes classifier. Paper presented at the Proceedings of the 3rd International Symposium on Big Data and Cloud Computing Challenges (ISBCC–16’).
Tian, C., Li, C., Zhang, G., & Lv, Y. (2019). Data driven parallel prediction of building energy consumption using generative adversarial nets. Energy and Buildings, 186, 230-243.
Tukey, J. W. (1977). Exploratory data analysis: Reading (MA): Addison-Wesley.
Uğuz, H. (2011). A two-stage feature selection method for text categorization by using information gain, principal component analysis and genetic algorithm. Knowledge-Based Systems, 24(7), 1024-1032.
Wang, W., Wang, C., Cui, T., & Li, Y. (2020). Study of Restrained Network Structures for Wasserstein Generative Adversarial Networks (WGANs) on Numeric Data Augmentation. IEEE Access, 8, 89812-89821.
Whitney, A. W. (1971). A direct method of nonparametric measurement selection. IEEE Transactions on Computers, 100(9), 1100-1103.
Yang, Y., & Pedersen, J. O. (1997, July). A comparative study on feature selection in text categorization. In Icml (Vol. 97, No. 412-420, p. 35).
Zadeh, L. A. (1965). Fuzzy sets. Information and Control, 8(3), 338-353.
Zhao, B., Chang, B., Jie, Z., & Sigal, L. (2018). Modular generative adversarial networks. In Proceedings of the European conference on computer vision (ECCV) (pp. 150-165).
Zhou, J., Duan, B., Huang, J., & Li, N. (2015). Incorporating prior knowledge and multi-kernel into linear programming support vector regression. Soft Computing, 19(7), 2047-2061.